2 March 2012 High-accuracy background model for real-time video foreground object detection
Author Affiliations +
Optical Engineering, 51(2), 027202 (2012). doi:10.1117/1.OE.51.2.027202
Abstract
Video foreground object detection faces the problems of moving backgrounds, illumination changes, chaotic motion in real word applications. This paper presents a hybrid pixel-based background (HPB) model, which is constructed by single stable record and multi-layer astable records after initial learning. This HPB model can be used for background subtraction to extract objects precisely in various complex scenes. Using the multi-layer astable records, we also propose a homogeneous background subtraction that can detect the foreground object with less memory load. Based on the benchmark videos, the experimental results show that single stable and 3-layer multi-layer astable records can be enough for background model construction and are updated quickly to overcome the background variation. The proposed approach can improve the average error rates of foreground object detection up to 86% when comparing with the latest works. Furthermore, our method can achieve real-time analysis for complex scenes on personal computers and embedded platforms.
Tsai, Lin, and Sheu: High-accuracy background model for real-time video foreground object detection

1.

Introduction

Foreground object detection in a complex scene is an important step in many computer vision applications, such as visual surveillance,12.3 intelligent behavior recognition45.6.7 and vehicle motion tracking.89.10.11.12 It is always desirable to accurately extract the foreground object(s). These applications are highly dependent on the results of foreground object detection. A common and efficient approach for extracting foreground objects from complex background is using background subtraction methods.1314.15.16 Background subtraction methods detect foreground objects by comparing each current frame with its background model. A difficulty with the background subtraction method is that complex scenes are usually dynamic. Complex scenes could be caused by waving trees, falling rain, illumination changes, and other background changes. To process these complex scenes, the technology of a robust background model is crucial.

Early background models17181920 had the advantage of low memory consumption and high processing speed. These approaches work well with stationary scenes, but usually they cannot handle complex scenes properly. Therefore, a number of background modeling methods have been proposed, and the most common one is the Mixture of Gaussians Model (MoG).21 By using more than one Gaussian distribution per pixel, it is able to handle complex scene, but MoG consumes more memory space and processing time. Therefore, some methods2223.24 proposed to improve this drawback. Instead of MoG, Kim13 presented a real-time object detection algorithm based on codebook model that is efficient in either memory or processing time, but it does not take into account dependence between adjacent pixels. Heikkila and Pietikainen25 presented a novel approach to background subtraction, in which the background is modeled by texture features. It is capable of real-time processing under 160×120 image size. Wang3 presented a background modeling method called SACON that computes sample consensus and estimates a statistical model of the background scene. Chan and Chien26 used a multi-background registration technique to calculate weight value for each pixel to update their background model. According to the weight value, the updating mechanism determines whether the pixel is replaced or not. thus it consumes less memory and computation time. Chiu27 proposed a fast background model construct algorithm that improved the original weighting average method. It uses a probability-based algorithm to construct the background model and to detect the object. This approach works well with slight background changes, but it needs connect-component labeling methods to overcome the challenges of complex scenes.

In recent years, video compression technology and neural networks have been used to solve many problems in video surveillance. Wang28 proposed a background modeling approach in the frequency domain, which constructs background model using DCT coefficients to achieve lower processing time. However, it has difficulty handling complex scenes. Maddalean and Petrosino29 presented a new self-organizing method for modeling background by learning motion patterns that was employed to model complex scenes. This method records the weight vector at each pixel by using a large memory to store the neuronal map. Tsai30 proposed a Fourier spectrum background model that can adapt to illumination changes, but it is only suitable for indoor work and grayscale images. Ref. 31 proposed a background subtraction algorithm called Vibe that can be initialized with a single frame. This algorithm was embedded in a digital camera which has a low speed ARM processor. Ref. 32 presents a real-time approach for background subtraction which can overcome gradual and sudden illumination changes. This approach segments each pixel by using a probabilistic mixture-based and non-parametric model.

In this paper, a hybrid background modeling approach is based on stable and multi-layer astable records. It can effectively offer foreground object detection in complex scenes. Our approach is applicable to those background pixels which vary over time. In the detection phase, it takes into account the dependence of adjacent pixels in the astable background record by using homogeneous background subtraction; therefore, we can extract the foreground object with a low error rate. In this way, our hybrid pixel-based background (HPB) model and detection method are resistant against erroneous detection in the complex scene.

2.

Proposed HPB Model and Homogenous Background Subtraction

In our proposed foreground object detection system, the block diagram is illustrated in Fig. 1, which includes learning and detection phases. In the beginning, the video input is switched to the learning phase. According to the pixel variation and analysis, the steady pixels are kept in a stable record and the varying pixels are saved by the multi-layer astable records. After using the above records to construct an HPB model completely, the system switches the input sequence into the detection phase and starts to perform the foreground object detection.

Fig. 1

The block diagram of the proposed foreground object detection.

OE_51_2_027202_f001.png

2.1.

Creating a Stable Background Record

In the learning phase, stable pixel analysis is applied to make a stable background record (SBR). For a video sequence with n×n-pixel frame size, let xi,j(t) be denoted as the pixel at location (i,j) in RGB at time t, as shown in Eq. (1). Equation (5) is used to assess the similarity between two pixel xi,j(t) and xi,j(t-1). These two pixels are regarded as similar if their difference is smaller than threshold value (Th=30). Stable Time, STi,j(t), is used to store the duration of those pixels that are alike; a value of STi,j(t) stands for the pixel has unchanged for a period of time. If xi,j(t) is not similar to xi,j(t-1), Stable Time will be reset to 0 and counting is resumed. Thus, the value of STi,j(t) implies the stability of a pixel, and stable background record can be built according to Eqs. (1) and (2),

(1)

xi,j(t)=[Ri,j(t),Gi,j(t),Bi,j(t)],

(2)

STi,j(t)={STi,j(t1)+1if|xi,j(t)xi,j(t1)|Th0,else,
where Th is a threshold for evaluating the similarity. Thus, a stable background can be built when the value of STi,j(t) is large enough. The stable background record (SBRi,j) can be obtained according to Eq. (3).

(3)

SBRi,j={xi,j(t),ifSTi,j(t)Th_Sϕ,else.
The SBRi,j is a empty set when the stable time STi,j(t)of a pixel is less than Th_S. The pixel is stable for a long period of time if STi,j(t) is larger than or equal to Th_S. Thus, the SBRi,j can obtain the value of xi,j(t).

2.2.

Constructing the Astable Background Record

Building the astable background record (ABR) is important since a stable background record is not sufficient to obtain a foreground object precisely. The astable background record consists of multi-layer 2-dimension buffers to store the complex scene. In the learning phase, the record ABRi,ju stands for one background pixel stored in (i,j) buffer of the uth layer buffer. Let MCi,ju represent the match counter to count how many xi,j(t) is matched to ABRi,ju. The multi-layer astable background record is gradually established frame by frame.

  • Step 1: Initialization at t=1;

    ABRi,j1=xi,j(1)

  • Step 2: Major learning for the pixel of frames from t=2 to N;

    for i,j=1,2,,n

    • Step 2.1: Find xi,j(t) matching to the ABRi,ju, while ABRi,juΦ

      if |xi,j(t)ABRi,ju|Th

      then, MCi,ju=MCi,ju+1;

      end

    • Step 2.2: If there is no match, save the input pixel into new layer ABRi,ju.

      ABRi,ju=xi,j(t) end

  • Step 3: Release the ABRi,ju based on the criteria (a) or (b)

    (a)MCi,ju<Th_f

    (b)|ABRi,juSBRi,ju|<Th

In the beginning frame, the input pixel xi,j(1) is stored in the first layer of the ABRi,j1. For the second frame, the pixel xi,j(2) is compared with the corresponding ABRi,j1. If they are similar, then MCi,j1 will be increased. If there is no match, then xi,j(2) will be stored in the second layer of ABRi,j2, and so on. At the end of learning phase, we have to delete the useless ABRi,ju which meet criteria (a) or (b) to reduce the memory requirements. When MCi,ju is less than Th_f(Th_f=15), it means that this ABRi,ju could be a foreground element appearing temporarily or noise. If ABRi,ju is similar to SBRi,j, the corresponding ABRi,ju should be deleted to save memory space.

Figure 2 shows an example of finding a stable background record (SBR) and 3-layer ABR. Figure 2(a) illustrates a static background record after 300 learning frames. Figure 2(b) shows the ABR1, in which shaking leaves, falling rain, and water were recorded correctly as an astable background. Figures 2(c) and 2(d), which show the other shaking leaves, falling rain, are the second and the third layers of the astable background record, respectively.

Fig. 2

HPB model at 300th frame: The stable background record (a), and the astable backgrounds ABR1 (b), ABR2 (c), and ABR3 (d) for the dynamic scene.

OE_51_2_027202_f002.png

2.3.

Foreground Object Detection with Homogenous Background Subtraction

After the construction of HPB model, foreground objects can be obtained by background subtraction. However, according to Figs. 2(b)2(d), it can be observed that the astable background records have composition of homogeneous blob movements in the shaky area. In order to reduce detection error and save recording memory, the characteristic of homogeneity in an area has to be taken into account while performing the background subtraction. Thus, the input xi,j(t) is compared to the neighbors ABRi±s,j±pu in area ((s,p1,2,r). The neighboring area can be r×rpixels and centered at xi,j(t). A foreground object (FO) can be detected by homogenous background subtraction, as in Eq. (4):

(4)

FOi,j(t)={0,if|xi,j(t)SBRi,j|Th0,if|xi,j(t)ABRi±s,j±pu|Th1,else,
where s,p=1,2,,r and u=1,2,,m. While FO is equal to 1, it stands for foreground object pixel; otherwise, it represents background.

3.

Background Updating

In the detection phase, we must update the HPB model over time to prevent detection errors resulting from outdated background information. Since the SBR and ABR are established in different ways, methods for updating them are different as well. The SBR is composed of stable pixels, therefore only background information updates are needed. On the other hand, updating the background information for ABRs needs a more complex replacement mechanism.

In Eq. (5), when SBRi,j and xi,j(t) match, we use the running average to update the corresponding SBRi,j,

(5)

SBRi,j=α×xi,j(t)+(1α)×SBRi,j,if|SBRi,jxi,j(t)|<Th,
where α is a constant whose value is less than 1.

Similarly, as in Eq. (6), when xi,j(t) and ABRi,ju match, we also use the running average to update the ABR. However, when xi,j(t) and ABRi,ju do not match, we use an exponential distribution probability density function to determine if ABRi,ju should be replaced. Thus, pri,ju represents the probability of whether or not the replacement should occur. The ABR update and replacement procedure is as follows: Compare the input pixel xi,j(t) with ABRi,ju for all i,j,u.

  • Step 1: If xi,j(t) matches to ABRi,ju(|ABRi,juxi,j(t)|<Th).

    (6)

    ABRi,ju=α×xi,j(t)+(1α)×ABRi,j

  • Step 2: If xi,j(t) not match to ABRi,ju.

    • Step 2.1: Find the minimum probability value of ABRi,ju.

      k=argmin{pri,ju|u=1,2,3,L,m}

    • Step 2.2: Delete the minor probability value in ABRi,ju.

      ifpri,ju<Th_p

      ABRi,jk=xi,j(t)

      end

where Th_p is a threshold for probability. And then, the probability value can be obtained as

(7)

pri,ju(t)=MCi,juN×exp(MCi,juN×TIi,ju(t))u=1,2,,m,
where MCi,ju is obtained in learning phase and N is the total number of learning frames. If xi,j(t) does not match to ABRi,ju, the TIi,ju(t) will be increased by one, and then the pri,ju will decrease exponentially. TIi,ju(t) is defined as a counter that records the time interval of ABRi,j matching to xi,j(t). It can be obtained as in Eq. (8),

(8)

TIi,ju(t)={TIi,ju(t1)+1,if|xi,j(t)ABRi,ju|0,elseu=1,2,,m,
where TIi,ju(0)=0. Thus, a large value of TIi,ju(t) means that xi,j(t) and ABRi,ju are not matched for a long period of time. It is resumed while xi,j(t) and ABGi,ju are good match.

4.

Experimental Results and Comparison

To evaluate the performance of background subtraction, three test video sequences including waving trees, torrential rain and wind, and PETS’2001 Dataset 333 were used in the experiments. The performance of the proposed method was compared with that of Codebook,13 MOG (Wu),23 SACON,3 Chien,26 and ViBe.31 A pixel-based error rate based on ground truth is a fair and often adapted assessment method,27 and was used to evaluate each method’s performance. The error rate is given in Eq. (9),

(9)

Error Rate=fp+fnframe size,
where fp and fn are the sum of all false positives and the sum of false negatives, respectively. A smaller error rate means that the detected result is more similar to the ground truth.

It is important to choose a proper number of background layers to trade off between the hardware memory and the number of ABR layers based on the scene requirements. Figures 3(c)3(g) show the foreground detection results for the waving trees video images, based on our proposed HPB model with 1 to 5-layer ABRs. In Fig. 4, the error rates of the 3-layer, 4-layer, and 5-layer ABRs are small. However, the 3-layer ABR uses much less memory.

Fig. 3

Foreground detection results of 1-, 2-, 3-, 4-, and 5-layer ABR: (a) source image, (b) ground truth, (c) 5-layer ABR, (d) 4-layer ABR, (e) 3-layer ABR, (f) 2-layer ABR, (g) 1-layer ABR.

OE_51_2_027202_f003.png

Fig. 4

Error rates of 1-, 2-, 3-, 4-, and 5-layer ABRs.

OE_51_2_027202_f004.png

In Figs. 5Fig. 67, we demonstrate that the proposed approach exhibits better foreground detection than the other methods in three benchmarks. Furthermore, Figs. 8Fig. 910 demonstrate that the proposed method presents a lower error rate in the ground truth comparison. The average error rates of the results from the six methods for various sequences are depicted in Table 1. It shows that the proposed approach has lower average error rate than the other methods.

Fig. 5

Object detection results in a moving background: (a) source image, (b) ground truth, (c) Codebook, (d) MoG (Wu), (e) SACON, (f) Chien, (g) ViBe, (h) Proposed.

OE_51_2_027202_f005.png

Fig. 6

Object detection results with illumination changes: (a) source image, (b) ground truth, (c) Codebook, (d) MoG (Wu), (e) SACON, (f) Chien, (g) ViBe, (h) Proposed.

OE_51_2_027202_f006.png

Fig. 7

Object detection in the raining sequence: (a) source image, (b) ground truth, (c) Codebook, (d) MoG (Wu), (e) SACON, (f) Chien, (g) ViBe, (h) Proposed.

OE_51_2_027202_f007.png

Fig. 8

Error rate in the waving tree sequence.

OE_51_2_027202_f008.png

Fig. 9

Error rate in the PETS’2001 sequence.

OE_51_2_027202_f009.png

Fig. 10

Error rate in the raining sequence.

OE_51_2_027202_f010.png

Table 1

Average of error rate in various benchmarks.

Video sequenceCodebookMoG (Wu)SACONChienViBe (N=20)Proposed
Waving tree(160×120)4.11%2.42%2.93%7.75%4.41%1.05%
Raining(320×240)2.49%1.56%1.31%3.19%2.14%0.81%
PETS 2001(768×576)0.93%1.73%0.92%1.37%0.78%0.75%

As shown in Fig. 11, we use the TI TMS320DM6446 Davinci development kit as our development platform which has a dual-core device including ARM926EJ-S and C64x+TM DSP. The resources of an embedded platform are limited, so the implementation has to consider memory consumption. Table 2 lists the real memory utilization for all six methods when applied to the different video sequences. It shows the memory requirement of our proposed method is much less than other approaches; thus, our approach can achieve the real-time operation with 23 frames per second for the waving trees video. The proposed method is suitable for implementation in an embedded platform.

Fig. 11

Implementation of the embedded system.

OE_51_2_027202_f011.png

Table 2

Memory comparison of background models.

Video sequenceCodebookMoG(Wu)SACONChienViBe(N=20)Proposed
Waving tree(160×120)1638 KB1920 KB1459 KB1094 KB1152 KB1056 KB
Raining(320×240)5330 KB7680 KB5836 KB4377 KB4608 KB4224 KB
PETS 2001(768×576)26.5 MB44.2 MB33.6 MB25.2 MB26.5 MB24.3 MB

5.

Conclusions

An efficient and precise foreground object detection method was proposed in this paper. The proposed method applies a stable background record and multi-layer astable background records to construct a correct background model. While when more layers are used, more background information can be recorded to improve the precision, it also needs more memory as well as more calculation effort. Thus, it is important to choose a proper number of background layers to trade off between the memory load and the number of dynamic background layers required by the scene. To save more memory space and calculation time, the 3-layer dynamic background model was used in our approach. According to our experimental results, the error rates of the 3-layer, 4-layer, and 5-layer HPB models are similar in many video benchmarks. The results demonstrate that the proposed method has a lower error rate with ground truth than the five other models tested. Furthermore, the proposed approach has higher precision of object detection than other methods for various sequences. The final verification was done using a 2.66 GHz CPU with a video resolution of 768×576 and an execution speed of 21frames/second in complex background scene. In addition, the proposed method can achieve real-time for complex scene on a Davinci embedded platform.

Acknowledgments

We would like to thank Oliver Barnich and M. Van Droogenbroeck, who provided the C-like source code for his algorithms.

References

1. 

C. StaufferW. Grimson, “Adaptive background mixture models for real-time tracking,” in Proc. IEEE Conf. Computer Vision and Pattern Recogn., Vol. 2, pp. 246–252 (1999).Google Scholar

2. 

A. CavallaroO. SteigerT. Ebrahimi, “Tracking video objects in cluttered background,” IEEE Trans. Circuits Syst. Video Technol. 15(4), 575–584 (2005).ITCTEM1051-8215http://dx.doi.org/10.1109/TCSVT.2005.844447Google Scholar

3. 

H. WangD. Suter, “A consensus-based method for tracking modeling background scenario and foreground appearance,” Pattern Recogn. 40(3), 1091–1105 (2006).PTNRA80031-3203http://dx.doi.org/10.1016/j.patcog.2006.05.024Google Scholar

4. 

J. W. Hsiehet al., “Video-based human movement analysis and its application to surveillance systems,” IEEE Trans. Multimedia 10(3), 372–384 (2008).ITMUF81520-9210http://dx.doi.org/10.1109/TMM.2008.917403Google Scholar

5. 

C. F. Juanget al., “Computer vision-based human body segmentation and posture estimation,” IEEE Trans. Syst. Man Cybernet. A: Syst. Hum. 39(1), 119–133 (2009).1083-4427 http://dx.doi.org/10.1109/TSMCA.2009.2008397Google Scholar

6. 

W. LaoJ. HanP. H. N. de With, “Automatic video-based human motion analyzer for consumer surveillance system,” IEEE Trans. Consumer Electron. 55(22), 591–598 (2009).ITCEDA0098-3063http://dx.doi.org/10.1109/TCE.2009.5174427Google Scholar

7. 

C. H. Chuanget al., “Carried object detection using ration histogram and its application to suspicious event analysis,” IEEE Trans. Circuits Syst. Video Technol. 19(6), 911–916 (2009).ITCTEM1051-8215http://dx.doi.org/10.1109/TCSVT.2009.2017415Google Scholar

8. 

S. Gupteet al., “Detection and classification of vehicles,” IEEE Trans. Intel. Transport. Syst. 3(1), 37–47 (2002).1524-9050http://dx.doi.org/10.1109/6979.994794Google Scholar

9. 

S. C. Chenet al., “Learning-based spatio-temporal vehicle tracking and indexing for transportation multimedia database systems,” IEEE Trans. Intel. Transport. Syst. 4(3), 154–167 (2003).1524-9050http://dx.doi.org/10.1109/TITS.2003.821290Google Scholar

10. 

J. W. Hsiehet al., “Automatic traffic surveillance system for vehicle tracking and classification,” IEEE Trans. Intel. Transport. Syst. 7(2), 175–187 (2006).1524-9050http://dx.doi.org/10.1109/TITS.2006.874722Google Scholar

11. 

W. ZhangQ. M. Jonathan Wu, “Moving vehicles detection based on adaptive motion histogram,” Digital Signal Process. 20(3), 793–805 (2010).DSPREJ1051-2004http://dx.doi.org/10.1016/j.dsp.2009.10.006Google Scholar

12. 

N. BuchJ. OrwellS. A. Velastin, “Detection and classification of vehicles for urban traffic scenes,” in Int. Conf. Vis. Inform. Eng., pp. 182–187 (2008).Google Scholar

13. 

K. Kimet al., “Real-time foreground-background segmentation using codebook model,” Real-Time Imag. 11(3), 172–185 (2005).http://dx.doi.org/10.1016/j.rti.2004.12.004Google Scholar

14. 

P. Spagnoloet al., “Moving object segmentation by background subtraction and temporal analysis,” Image Vis. Comput. 24(5), 411–423 (2006).IVCODK0262-8856http://dx.doi.org/10.1016/j.imavis.2006.01.001Google Scholar

15. 

J. C. NascimentoJ. S. Marques, “Performance evaluation of object detection algorithms for video surveillance,” IEEE Trans. Multimedia 8(4), 761–774 (2006).ITMUF81520-9210http://dx.doi.org/10.1109/TMM.2006.876287Google Scholar

16. 

P. M. JodoinM. MignotteJ. Konrad, “Statistical background subtraction using spatial cues,” IEEE Trans. Circuits Syst. Video Technol. 17(12), 1758–1763 (2007).ITCTEM1051-8215http://dx.doi.org/10.1109/TCSVT.2007.906935Google Scholar

17. 

C. R. Wrenet al., “Pifnder: real-time tracking of the human body,” IEEE Trans. Pattern Anal. Mach. Intel. 19(7), 780–785 (1997).ITPIDJ0162-8828http://dx.doi.org/10.1109/34.598236Google Scholar

18. 

I. HaritaogluD. HarwoodL. S. Davis, “W4: real-time surveillance of people and their activities,” IEEE Trans. Pattern Anal. Mach. Intel. 22(8), 809–830 (2000).ITPIDJ0162-8828http://dx.doi.org/10.1109/34.868683Google Scholar

19. 

C. KimJ. N. Hwang, “Object-based video abstraction for video surveillance systems,” IEEE Trans. Circuits Syst. Video Technol. 12(12), 1128–1138 (2002).ITCTEM1051-8215http://dx.doi.org/10.1109/TCSVT.2002.806813Google Scholar

20. 

S. Y. ChienS. Y. MaL. G. Chen, “Efficient moving object segmentation algorithm using background registration technique,” IEEE Trans. Circuits Syst. Video Technol. 12(7), 577–586 (2002).ITCTEM1051-8215http://dx.doi.org/10.1109/TCSVT.2002.800516Google Scholar

21. 

C. StaufferW. Grimson, “Learning patterns of activity using real-time tracking,” IEEE Trans. Pattern Anal. Mach. Intel. 22(8), 747–757 (2000).ITPIDJ0162-8828http://dx.doi.org/10.1109/34.868677Google Scholar

22. 

D. S. Lee, “Effective Gaussian mixture learning for video background subtraction,” IEEE Trans. Pattern Anal. Mach. Intel. 27(5), 827–832 (2005).ITPIDJ0162-8828http://dx.doi.org/10.1109/TPAMI.2005.102Google Scholar

23. 

H. H. P. Wuet al., “Improved moving object segmentation by multi-resolution and variable thresholding,” Opt. Eng. 45(11), 117003 (2006).1389-4420http://dx.doi.org/10.1117/1.2393227Google Scholar

24. 

J. Chenget al., “Flexible background mixture models for foreground segmentation,” Image Vis. Comput. 24(5), 473–482 (2006).IVCODK0262-8856http://dx.doi.org/10.1016/j.imavis.2006.01.018Google Scholar

25. 

M. HeikkilaM. Pietikainen, “A texture-based method for modeling the background and detecting moving objects,” IEEE Trans. Pattern Anal. Mach. Intel. 28(4), 657–662 (2006).ITPIDJ0162-8828http://dx.doi.org/10.1109/TPAMI.2006.68Google Scholar

26. 

W. K. Chanet al., “Efficient content analysis engine for visual surveillance network,” IEEE Trans. Circuits Syst. Video Technol. 19(5), 693–703 (2009).ITCTEM1051-8215http://dx.doi.org/10.1109/TCSVT.2009.2017408Google Scholar

27. 

C. C. ChiuM. Y. KuL. W. Liang, “A robust object segmentation system using a probability-based background extraction algorithm,” IEEE Trans. Circuits Syst. Video Technol. 20(4), 518–528 (2010).ITCTEM1051-8215http://dx.doi.org/10.1109/TCSVT.2009.2035843Google Scholar

28. 

W. WangJ. YangW. Gao, “Modeling background and segmenting moving objects form compressed video,” IEEE Trans. Circuits Syst. Video Technol. 18(5), 670–681 (2008).ITCTEM1051-8215http://dx.doi.org/10.1109/TCSVT.2008.918800Google Scholar

29. 

L. MaddaleanA. Petrosino, “A self-organizing approach to background subtraction for visual surveillance applications,” IEEE Trans. Image Process. 17(7), 1168–1177 (2008).IIPRE41057-7149http://dx.doi.org/10.1109/TIP.2008.924285Google Scholar

30. 

D. M. TsaiW. Y. Chiu, “Motion detection using Fourier image reconstruction,” Pattern Recogn. Lett. 29(16), 2145–2155 (2008).PRLEDG0167-8655http://dx.doi.org/10.1016/j.patrec.2008.08.005Google Scholar

31. 

O. BranichM. Van Droogenbreck, “ ViBe : a universal background subtraction algorithm for video sequences,” IEEE Trans. Image Process. 20(6), 1709–1724 (2011).IIPRE41057-7149http://dx.doi.org/10.1109/TIP.2010.2101613Google Scholar

32. 

A. LanzaS. SaltiL. Di Stefano, “ Background subtraction by non-parametric probabilistic clustering,” in IEEE Int. Conf. Adv. Video Signal-Based Surveil., IEEE Society, Klagenfurt, pp. 243–248 (2011).Google Scholar

33. 

PETS’2001 Dataset 3. http://www.cvg.cs.rdg.ac.uk/PETS2001/pets2001-dataset.html Last modified 6 July (2009).Google Scholar

Biography

OE_51_2_027202_d001.png

Wen-Kai Tsai is a PhD candidate at Graduate School of Engineering Science and Technology, National Yunlin University of Science and Technology, Taiwan. He received BS and MS degrees in electronics engineering from National Yunlin University of Science and Technology, Taiwan, in 2004 and 2006, respectively. His research interests include digital signal processing and image processing.

OE_51_2_027202_d002.png

Chung-Chi Lin received the MS degree in computer science from University of Houston, TX, USA, in 1983, and the PhD degree in engineering science and technology from National Yunlin University of Science & Technology, Taiwan, in 2009. Since 2009, he has been an Associate Professor with the Department of Computer Science, Tunghai University, Taiwan. His current research interests include image processing, digital signal processing, and System-on-chip design.

OE_51_2_027202_d003.png

Ming-Hwa Sheu received the BS degree in electronics engineering from National Taiwan University of Science and Technology, Taipei, Taiwan, in 1986, and the MS and PhD degrees in Electrical Engineering from National Cheng Kung University, Tainan, Taiwan, in 1989 and 1993, respectively. From 1994 to 2003, he was an Associate Professor at National Yunlin University of Science and Technology, Touliu, Taiwan. Currently, he is a Professor in the Department of Electronics Engineering, National Yunlin University of Science and Technology. His research interests include CAD/VLSI, digital signal processing, algorithm analysis, and system-on-chip design.

Wen-Kai Tsai, Ming-Hwa Sheu, Chung-Chi Lin, "High-accuracy background model for real-time video foreground object detection," Optical Engineering 51(2), 027202 (2 March 2012). http://dx.doi.org/10.1117/1.OE.51.2.027202
Submission: Received ; Accepted
JOURNAL ARTICLE
10 PAGES


SHARE
KEYWORDS
Video

Performance modeling

Motion models

Optical engineering

RGB color model

Video surveillance

Positron emission tomography

RELATED CONTENT


Back to Top